Genetic fine-mapping from summary data using a nonlocal prior improves the detection of multiple causal variants

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) identify genomic loci linked to complex traits.
• Genetic fine-mapping refines these loci to pinpoint independent causal variants, accounting for linkage disequilibrium.

Purpose of the Study:

• To introduce FiniMOM, a novel Bayesian fine-mapping method for summarized genetic association data.
• To improve the accuracy and power of identifying causal variants in complex genetic loci.

Main Methods:

• FiniMOM utilizes a nonlocal inverse-moment prior for causal effects, suitable for finite samples.
• A beta-binomial prior models the number of causal variants, allowing control for linkage disequilibrium reference misspecification.
• The method was evaluated using simulation studies mimicking GWAS on circulating protein levels.

Main Results:

• FiniMOM demonstrated improved credible set coverage compared to the state-of-the-art SuSiE method.
• The proposed method showed enhanced power in detecting causal variants, especially in scenarios with multiple causal variants per locus.
• Simulation results confirmed FiniMOM's effectiveness in fine-mapping genetic loci.

Conclusions:

• FiniMOM offers a robust Bayesian approach for genetic fine-mapping using summarized association data.
• The method provides superior performance over existing tools, particularly for complex loci with multiple causal variants.
• FiniMOM is a valuable tool for researchers aiming to precisely identify causal variants from GWAS.